Bishop Hill Cook's progress
Sep 21, 2013 Climate: Sceptics Climate: other I have an opinion piece in Canada's Financial Post, taking a look at the global warming consensus as revealed in a series of studies, including the Cook one.
Once the methodology used by Cook and his colleagues is understood, it becomes abundantly clear that the consensus it describes is a very shallow one; the results add up to little more than “carbon dioxide is a greenhouse gas” and “mankind affects the climate.” These are propositions that almost everybody in the climate debate accepts; the argument continues to be over how much greenhouse gases have affected us in the past and how much they will affect us in the future, and whether any of this represents a problem.
By coincidence, John Cook has been given the right of reply to my piece last week in the Australian looking at the same question. There's a lot of huffing and puffing, but I don't think he nails it.
Reader Comments (59)
Richard B,
Thanks for your replies here and on the IPCC thread.
Looking at the data you linked to earlier, the ecosystem and oceans appear to be absorbing an increasing proportion of the excess CO2. There must be some lag in this response - at least one growing season for plants - so this "negative feedback" seems likely to strengthen with time.
I need to do some background reading on CO2 - the history of measurement, effects on plants, etc. - to get a better understanding. My present - limited - understanding is that records of CO2 levels over the recent past have a bit of a "hockey-stick" profile which has unfortunate connotations, shall we say.
(To be read, or internally heard, in best Cap'n Bob - as played by Tom Baker- old seadog manner.)
Why 'tis a mighty thin, modest and shy, homeopathic kind of consensus as I sees it.
That there ought to be some sort of a strategem apart, the technical criticism of Cook's paper (which has only been made at some length by Tol, Shollenberger, and yours truly, AFAICT) is directly related to an overall approach as taken by the Bish and/or Monckton. The technical approach simply quantitates the impact of the improper Cook method; it provides proof for what seems intuitively evident.
Anyone faced with task of classifying 12,000 abstracts will go down the same path. You wouldn't want to miss papers that support the consensus (assuming such things were happening in abstracts) by setting your bar too high. You wouldn't want (presumably) pull all kinds of papers in either.
[1] Cook's team settled for a sufficiently low threshold. This makes their test highly sensitive; but it dramatically bumps up the false positivity rate.
[2] How exactly does the Cook method end up as a filter for a 'shallow consensus'? The answer, surprisingly, is not in that these are activists ready to bump any paper into the 'consensus' category, but because of the inclusion of a 'implicit acceptance' category. This is evident from their own identification of a remarkably small number of hard consensus papers.
[3] Surprise of surprises, searching for 'global warming' does not bring up papers in academic literature that pertain to global warming. 'Global warming' is a non-technical term, used by people in the impacts, mitigation and public health research settings. These are groups who've assumed global warming to be true, or are agnostic to its causation. The abstract text of their papers do not contain information about acceptance or rejection of 'global warming'. Incidentally, this point is made succinctly by Monckton. The Cook group assume the opposite to be true. They don't run tests. There is no indication they knew the structure of their search results.
[4] The implicit acceptance category depends on volunteer's interpretation. Given the kind of abstracts thrown by the chosen search terms, the application of a non-discriminant classification scale onto a mass of undifferentiated text bins abstracts into meaningless categories, regardless of any true consensus content. Given the above, raters would find it difficult to classify abstracts, as neither the classification system nor the abstract content provides good handles to hold on to. Classification skewness and drift and evidence for fatigue would be evident. Neutral categories would be difficult to reproducibly be distinguished from the 'implicit' categories. These are shown by Tol's analysis. If Cook releases the full data, the details can be demonstrated (or not) fully.
[5] Cook's method of 'confirmation' of their result is the worst. It is an absolute, unqualified fraud. The thousand or so papers classified by authors, and their abstracts classified by Cook's team show ratings that are significantly discordant. Cook calculates a final '97%' figure and flies with it. It is incredible. It is like going to the zoo with children in the van and coming home with a different set of kids and claiming everything's alright because the 'total headcount is ok'.
[6] The utter shambles the classification system is, is shown by the abysmal inter-rater reproducibility of ratings, and the lack of rater-vs-author reproducibility. Unexpectedly, the error in classification is not merely confined to ratings '3' and '4' though the majority is, but extends to other consensus categories as well.
Frank.
You reckon sea-level rise is a linear function of temperature rise? ('over millennia'). And sea level has gone up about 24 metres per degree rise? As I understand it, global temperature has gone up about a degree since the mid-19th century, and sea-level (maybe) less than a metre. That doesn't tie up.
Why should sea-level rise be directly proportional to temperature? In the ice ages, enormous quantities of ice covered the continents. Melting most of this would produce quite a lot of water. But after most of it has melted, there's no reason to suppose that further temperature rise would produce sea-level rise at the same rate (also, the melted icewater is not being collected in a cylindrical vessel, but in one with diverging sides). So your argument makes no sense to me. But feel free to correct my figures, assumptions and logic!
@Sep 22, 2013 at 7:50 PM shub
I withdraw that "shouldn't" bad phrasing but pretty presumptuous to say even if so :)
Of course not for me to say, personally I'd like to see the critique brought to bear across the spectrum, but I am staying at the level of your [3]. I agree with all that, and putting on a dreadful Wardian PR head, I think that, like a Pareto distribution, this does 80% of the work for 20% of the effort. All up to there can easily be shown to the layman, and nor do I see how this can be easily argued against -if it could be forced as the main issues.
I have not properly engaged beyond your [3] to be a critic of anything further. I have cast my eye over it and waited for something to gel but I am l too lazy to spend much more time on it. My technical analysis so far rises nowhere above the level of adding and dividing :) Just on that level was enough to be genuinely bemused by what I thought where technically able fans being silly.
I think the possibility of deriving some information on the rater behaviour from the remaining data is intriguing but I am passively waiting on that. I remember reading the treehut files and being struck how often Cook gave a constant refrain of emphasising that allowing papers to be consigned to the don’t know category was not a pressure. This knowledge of not effecting the result was something of a clue that they collectively knew they had so much to work with they could easily get their numbers. One of many subjective niggles I got when I went through the treehut files - after the prompting of the ridiculous hype this paper got!
I dunno I think there is stuff going on here regarding group think and poor self-discipline that could be shown technically, I am all ears ;)
Osseo: I certainly do not assume SLR is proportional to temperature. (The area of the earth shrinks as one approaches the poles, so SLR should shrink as melting approaches the poles.) My point was that even a small fraction of 24 m/degC poses a significant threat. Let's stop pretending that NO problem might exist and support responsible calculations for how much we are willing to pay today to avoid problems a century or more in the future. Richard Tol tells us we should pay very little, even if the IPCC's estimates of warming were right. Those estimate is probably 2-fold too high.
It does take a long time for warmer conditions to result in equilibrium SLR, possibly centuries. The end of the LIA was possibly responsible for SLR in the first half of the 20th century and it could still be continuing today - in addition to GHG driven SLR. It takes about 500 years some of the water at the bottom oceans to reach the surface and equilibrate with surface temperature.
Bishop (or someone else knowledgeable),
Could you please identify papers by prominent skeptics who were included in the 97% consensus in Cook's paper? Cook claimed that his methodology actually excluded skeptics and that for instance, he did not include Roy Spencer within the consensus camp. Cook does a lot of squirrelly things, and I would like to know if this is really one of them.
@Sep 23, 2013 at 2:55 PM | Unregistered CommenterJD Ohio
Sks have a method of searching by author the papers used in the paper on their site in this Search link:
As far as I know it is complete.
The search example for the author "Lindzen" in the link above shows Lindzen falls in both the endorse *and* minimises categories.
Thanks Leopard. I went to your link and inserted the last name, Lewis, hoping to find Nic Lewis. Didn't find him, but it was funny how off topic the articles found in the search were to the global warming topic.
JD